Frequency modulation schemes are well known in the art. One such scheme is a Gaussian frequency shift keying (GFSK) modulation where the digital information is transmitted through the corresponding variation of a discrete carrier frequency. For example, in a Bluetooth transmitter operating in an ideal environment absent noise and frequency offset, the transmit signal for a bit having a value of “1” has a frequency shift of +250 kHz (i.e., a positive shift) from the center frequency fc of the carrier, while the transmit signal for a bit having a value of “0” has a frequency shift of −250 kHz (i.e., a negative shift) from the center frequency fc of the carrier.
The receiver circuit in the above example is expecting the received modulated signal to be centered at the center frequency of the carrier for demodulation. However, mismatch between the transmitter and receiver may introduce a frequency offset error in the carrier frequency. As a result of this frequency offset error, the frequency demodulated waveform at the receiver exhibits in the baseband an average value for the frequency shifts which does not approach zero and is thus DC biased by a value corresponding to the magnitude of the frequency offset. For example, with a carrier frequency offset of +100 kHz, the frequency demodulated waveform at the receiver for the bit having the value of “1” will have a frequency shift of +350 kHz (from baseband) and the bit having the value of “0” will have a frequency shift of −150 kHz (from baseband). The DC bias in this example will have a positive voltage level.
The specification requirements for certain transceiver systems may permit some level of carrier frequency offset. For example, a known Bluetooth specification permits a carrier frequency offset of ±100 kHz. It is accordingly imperative for the receiver to include some form of carrier frequency offset compensation or removal technology that can handle the permitted ±100 kHz range in carrier frequency offset.
The Bluetooth link layer packet includes an eight-bit preamble field that is used by the receiver to perform a number of important tasks including frequency synchronization, symbol timing estimation, and automatic gain control (AGC) training. The preamble field is typically populated by a symbol (octet) formed of alternating logic bits (i.e., “10101010” or “01010101”). It is also known in the art to use the frequency demodulated preamble field data to detect a carrier frequency offset. Because of the alternating logic bits, the frequency demodulated waveform for the preamble, in the absence of a frequency carrier offset error, should exhibit in the baseband an average value of the frequency shifts that is substantially equal to zero (because there are an equal number of “1” and “0” bits in the symbol). If the average value is something other than zero, then it can be concluded that a frequency carrier offset error is present. The magnitude and sign of the corresponding DC bias in the frequency demodulated signal can be detected by filtering and is indicative of the magnitude and direction of the frequency carrier offset. The accuracy of this carrier frequency offset detection technique is improved if the analysis of the average value of the frequency shifts for the frequency demodulated waveform is performed over a plurality of preamble symbols.
As noted above, the preamble field is used for other purposes and this can introduce problems with achieving the carrier frequency offset detection. For example, the automatic gain control process must be performed prior to performing carrier frequency offset detection. To cater to the full range of the input radio frequency signal, multiple gains must be determined. The automatic gain control process itself can take the analysis of multiple demodulated preamble field symbols to complete, and the total length of time for this analysis may leave no room on the back side to perform the frequency carrier offset error analysis using multiple demodulated preamble fields. For low level signals, the automatic gain control process may be completed more quickly, but the latency introduced by the noise rejection process in digital filtering can be significant when a high number of blockers are considered and again little if any time may be left on the back side to perform the frequency carrier offset error analysis. In any event, it is noted that the fewer the number of frequency demodulated preamble field symbols that are made available for the frequency carrier offset error analysis, the lower the accuracy of the offset estimation.
There is accordingly a need in the art for an improved carrier frequency offset compensation technique that will address at least the foregoing problems and limitations of the conventional process for frequency carrier offset error analysis.